In some foreign country a priest, a lawyer, and a Six Sigma Black Belt are about to be guillotined. The priest puts his head on the block, they pull the rope, and nothing happens—he declares that he’s been saved by divine intervention—so he’s let go. The lawyer is put on the block, and again the rope doesn’t release the blade. He claims he can’t be executed twice for the same crime, and he is set free, too. They grab the Black Belt and shove his head into the guillotine. He looks up at the release mecha- nism and says, “Wait a minute, I see your problem . . .”
—Reprinted with permission, www.isixsigma.com
D
efects—no matter what the cause—can be stud- ied from several different angles, or points of view. But there is always a reason, and once we come to understand the reasons behind the defects, we can eliminate the frequency of those defects—not com- pletely, and not all the time, but enough so that the rate of defects will fall.Some Six Sigma processes involve the use of techni- cal language, but it really comes down to a fairly straight- forward process. In Six Sigma, we:
✔ Define a problem or series of problems, charac- terized by the defects that have been experienced in the past. (Define)
✔ Identify the causes of the defects. (Measure)
Chapter
✔ Estimate the cost of defects. (Analyze)
✔ Take steps to fix those defects. (Improve)
✔ Watch the results of our changes. (Control) The DMAIC process is the core of the Six Sigma working model, and whether you use highly technical language and graphics, or simply make checklists and fol- low the defects through the system to figure out how to remove them, the end result should be the same. For ex- ample, you might say that a particular defect involves
“cross functionality” or instead you might observe that a defect involves two or more departments.
For the purpose of understanding the concepts be- hind the Six Sigma process, we do not use terms that are not needed. The acronyms and abbreviations we employ in this book are widely recognized and used throughout the Six Sigma world; because they are universal, we have used them here as well. In your own company, the use of terms and acronyms will probably vary depending on the point of view of your Six Sigma leadership, the complexity of projects, and the size and scope of the work involved.
Key Point In Six Sigma projects, simplicity of termi- nology is preferred; in some cases, more complex terms and tools are justified, but not always.
You will be expected to confront two broad types of defects: tangible or intangible. A “tangible” defect is one that can be easily quantified. For example, on an assembly line, you can count the number of units produced in a shift, and you can identify exactly how many of those units were defective. If your standard is to have 5 percent or fewer defects, but they have been running between 15 percent and 20 percent, you have a problem. Under a Six Sigma project, it would not be adequate to say, “The shift supervisor has to make those employees pay more atten- tion” and then merely assume that will fix the problem.
You need to ask many other questions, such as:
✔ Are the raw materials adequate for this product?
✔ Have we changed materials or suppliers lately? Is the raw material different?
✔ Are employees properly trained?
✔ Is our machinery properly set? Can settings be adjusted and if so, does the adjustment level con- tribute to the defect?
✔ Is the 5 percent standard realistic? If not, what is a realistic standard?
✔ What other possible contributing factors are causing these defects?
In the tangible world, we can locate the causes of de- fects by asking these questions. The process is itself tangi- ble because we can easily count units of production and spot defects. It is specific. In comparison, an intangible product or service is much harder to measure. Both “units of production” (or, using the Six Sigma phrase, “opportu- nities”) and the number of defects may be far more diffi- cult to count and, as a result, more difficult to correct.
This is the challenge for anyone on a Six Sigma team in- volving intangible services and their defects.
MEASURING VARIANCE
The traditional measurement of defects, often expressed as variance, requires a reliable method for computing the normal expectations. Statisticians use a lot of formulation to arrive at what is called standard normal outcome. This is the normal distribution when the mean, or average, outcome is defined as zero, and a standard deviation is 1.
This is one of many ways to calculate and to assign a value to defects. A standard normal outcome would be the overall expectation, based on averages, of what you would be likely to experience in a series of tests. This type of analysis, popular in a manufacturing and production en- vironment where thousands of units are produced, has a valid application. In such an environment, a change of a small degree could mean a big difference in profitability through reduced waste, faster production time, accuracy, and other measurements. Cost accountants are also aware of the big differences that small changes make.
standard normal
the normal distri- bution of out- comes based on a statistical as- sumption that a mean is zero and a standard devia- tion is 1.
Key Point Attention to small details often yields big changes. If the nature of a project involves many opera- tions, the more attention to detail, the higher the yield.
Without requiring ourselves to become statisticians, engineers, or cost accountants, we can still devise accu- rate methods for measuring outcomes and for identifying ways to reduce defects. Six Sigma may have originated as a quality-related theory in a highly technical environ- ment, but it is desirable to make the techniques available to as many nontechnical people as possible.
If you work in a strictly production-related envi- ronment, you already know how defects are measured.
Units of production and defect-free outcome define a particular shift’s work level, or a particular line’s im- provement over previously established levels. Statisti- cians and accountants are obsessed with trends and changes, and they measure the success and effectiveness of change in terms of how those trends evolve. But it is not just the statistician or accountant; in one way or an- other, everyone is involved with measuring, evaluating, and controlling trends.
For example, if your work is strictly administrative, you have probably identified and defined a form of quality and you understand defects. A Word Processing Depart- ment, for example, may use several devices to ensure that the reports, letters, and other documents they process are fully accurate. Statistical reports have to be carefully dou- ble-checked to make sure no math errors go through.
Spell-checkers can be run against text documents and fur- ther edited to find errors in word usage, grammar, and formatting of documents. Many different skills and tech- niques can be employed to ensure that defects are held to a minimum.
In this environment, a “unit” may be defined as a single document or even as a page within a larger docu- ment. A defect could be defined as any error: spelling, word usage, grammar, math errors within the docu- ment, formatting inconsistencies, or even a smudge on the edge of the page. The definition of “defect” is going to depend on the precise nature of work that is
processed and this example—which is tangible but with a nonspecific unit count—defines the elusiveness of defining defect-free processes.
If we were to apply the strict standards of produc- tion, a “unit” would be each and every word within a doc- ument. In place of a word, a mathematical value or total would also represent a unit. Such detail, however, makes no sense in a statistical Word Processing Department or secretarial pool. The “unit of production” is probably the document and any defects in the document are important.
We would expect to create procedures to locate defects in draftform, so that these could be fixed before documents were released to the customer (executive, manager, super- visor, fellow employee, committee, board, etc.). Certainly, you would not want a math error to be found in the mid- dle of an accounting review meeting, or an obvious spelling error to be found by the recipient of a letter from your CEO. Procedures should be designed to ensure that (1) defects can be found in a consistent manner, (2) there is a process in place to correct defects, and (3) continual review is designed and implemented to keep the rate of defects as low as possible. We would be misguided to be- lieve that it would ever be possible to completely elimi- nate those defects; it is enough to find and correct them before output occurs.
Key Point A discovered defectcan and should be cor- rected during the process and in a proactive manner; and not after the output has occurred, in a reactive manner.
Even in a service or administrative department, as long as a process is predictable (meaning, of course, that you know exactly what occurs within that process) and as long as you know the types of defects, or errors, that are likely to occur, then you can determine what controls will be required to reduce defects. In the next chapter, the hor- izontal work flowchart is set up to demonstrate the types of control points where special care has to be given. For example, when a manager dictates a report and delivers it to a word processing pool, several possible defects can oc- cur. If the report includes grammatical errors, will the
word processor be able to catch it? Working from dicta- tion, it is possible to misunderstand and key in the wrong word. Consider a list of words or phrases that could be heard wrong or easily be mistyped:
Correct Word Heard As Correct Word Heard As a priori a priority listened listed
better bitter meant mint
costing causing nullify notify
dollar duller opinion open
east eased preference reference
familiar familial query clearly
guest guessed restitution resolution This list points out the need for careful editing. Spell checking is not enough. So here we have an example of potential defects that could easily occur in the weak link between a dictated letter and the processing of the tape. A second defect may occur at the proofreading stage where, instead of looking only for misspelled words, the individ- ual does not also readthe material to ensure that mistakes do not go through. So the sentence may read “Our client has asked for a summary of items causing this amount of overrun, and is familial with common reasons for such er- rors.” It should be corrected, replacing the word “causing”
with “costing” and replacing the word “familial” with “fa- miliar.” The first error may be difficult to find in all cases, so in order to entirely eliminate defects, we would depend on both states of the process to work well. It demon- strates, however, that we will not always be able to find the defects themselves. So all a Six Sigma project can do is to identify likely problems and do everything possible to eliminate them.
Are some defects impossible to measure? For ex- ample, going back to the word processing pool, what if the need to replace the word “causing” with “costing” is never found? If the word used accidentally conveys ap- proximately the same idea, does this count as a defect of the same level as a word change that completely
changes the sentence’s meaning? This is a difficult ques- tion, because every circumstance (like every example of word replacement) will contain subtle degrees of sever- ity. A word processing employee who makes a lot of mistakes and does not find or correct them presents a training problem, and requires double-checking by other employees as well. So the importance a particular employee attaches to error-free processing is an at- tribute that is difficult to measure. This makes it even more difficult to identify the likelihood of defects occur- ring within a department, when every employee may also apply different standards to their own work. What is acceptable? What is not?
One challenge for the Six Sigma team working in an environment with intangible processes, is to define expec- tations and requirements of the customer. So in a word processing pool, for example, the expectation might be for error-free documents, and the requirement may be that all documents are double-checked prior to release.
This may be the case for statistical reporting more than for text documents. For example, if the Accounting Depart- ment presents budgets, financial statements, and cost esti- mates to be prepared, these should be checked by at least two people in the department preparing the work, and a draft checked again by someone in the Accounting De- partment—all before final release.
The conflict often arises between the time needed to thoroughly execute work and deadlines. If a math-inten- sive document of many pages is presented for fast process- ing, and has to be completed within a few hours, that means (1) employees will feel pressured to work faster, meaning that (2) more errors are going to be made, and (3) less time will be available to double-check. So one de- fect in this process is the time pressure involved in how and when work has to be done. So a second challenge to the Six Sigma team in dealing with such defects is to iden- tify the causes beyond actual processing. In this example, input was late enough so that output was required within an unreasonable amount of time. As a result, either the work cannot be delivered as demanded, or it will be on time but probably with a greater frequency of errors. For a
financial report, any errors are going to be unacceptable.
The individual reviewing a financial report will have both an expectation and a requirement that the report be accu- rate. So it would be preferable for the report to be deliv- ered late, than it would be to make a quick deadline but issue the report with many errors.
This example demonstrates that pinning down a def- inition of a defect is not always going to be limited to the specific error that becomes obvious right away. A defect may be found in an unreasonable deadline; in poor qual- ity of information or raw material provided (a difficult to hear transcription, for example); or in the attitude of the poorly-trained employee. A lot of variances come into the picture, and a Six Sigma team should be able to consider the full range of possible problems in its analytical ap- proach to identifying and solving problems.
VARIANCE AND IMPROVEMENT TESTING Anyone who has never been involved with the actual test- ing of variances, knows that those variances themselves can be elusive, multifaceted, and ever-changing. It is rare to discover a variance that is (1) easily identified, (2) sta- tionary, and (3) easily remedied. If variances did contain these attributes, they would be easyto fix. We have to as- sume, as a starting point in the study of variances, that the easy ones are eliminated in due course, and that the Six Sigma team is assigned to tackle the particularly tough variances that remain chronically unchanged.
Key Point If finding and eliminating defects is easy, then those changes occur during the process. Six Sigma teamwork is needed when the problems are more complex and when they involve many people and departments.
Applying the scientific method to variance testing, we would have to begin with a basic hypothesis: What is the definition of the problemwe are supposed to address?In the example previously mentioned of the late deliveries coming from Shipping and Receiving, the initial belief is
that the shipping employees are not doing their jobs. In a methodical analysis of the problem, we discover that it is far more widespread and complex, and will require changes in many departments: Sales, Marketing, Account- ing, Inventory Control, and Shipping and Receiving. So in defining the problem, we discover the true scope and real- ize that it is not a simple one to fix. We cannot simply call the manager and say, “Send stuff out on time,” because the issues involve other process problems, and all of those have to be fixed before we can expect Shipping and Re- ceiving to do their job.
The next step in the testing process is, again, another question: What are we supposed to test?There are many el- ements to consider, referring to the SIPOC process map:
Suppliers, Input, Process, Output, and Customers. Any or all of these segments may involve areas requiring testing, and as part of the definition of what to test, each of these areas should be examined to find potential problems. Re- turning to the example of Shipping and Receiving, a num- ber of potential problem areas come up at each of these segments. For example:
Suppliers: Are materials being provided on time? If not, why not? How can this problem be corrected effec- tively and immediately? What internal processes have to change to facilitate timely delivery? (“Suppliers” may be defined as outside vendors for delivery of shipping sup- plies or the sales reps who need to deliver orders for ac- tion, so even this phase is complex in the definition itself.) Input: Is the Marketing Department processing orders in a timely manner? In fact, why do they have to be involved in processing between sales reps and Shipping/Receiving?
How can the process be improved for faster delivery of orders?
Process: What inventory problems make it difficult to fill orders? What is the cause of those problems? How can back order volume be reduced? How can inventory con- trols be improved? Do we even know what is in inventory right now? What changes have to be made in this system?
Output: What elements prevent prompt shipment in each point along the process? Since output involves sev- eral increments, where do we need to make changes?
What specific problems in Shipping and Receiving have to be changed? What changes have to be made in the Ac- counting Department to ensure that basic shipping sup- plies are on hand when needed?
Customers: What expectations are not being met?
What requirements are not being met? Is three-day deliv- ery a realistic goal? Do customers expect this? Would cus- tomers be just as happy with a seven-day promise of shipment, for example? Are we doing harm to ourselves by having sales reps make unrealistic commitments? If the three-day shipping promise is realistic, what list of changes are needed to ensure that we meet it all the time?
The third step in variance testing using the scientific method, is How will our proposed changes affect revenues and earnings?The cost element is an integral part of any customer service project. And given the Six Sigma philos- ophy, allprojects are related to customer service in some form or another. We need to perform an analysis of how projects increase revenues or reduce costs and expenses, so that earnings are improved. From the corporate and fi- nancial point of view, this is going to be the ultimate cen- ter of judgment. Appropriately so, the CEO and CFO of the organization and all managers in the reporting chain have questions on their minds at all times: How does a particular decision affect profits? Why is this appropriate?
If the company is in the business of earning profits for its shareholders, it is a responsibility of management to pursue ever-growing revenue volume and profits, ensuring that costs and expenses are held to a reasonable level and other- wise promoting the shareholders’ interests. With this in mind, it is everyone’s job within the company to become part of the “profit team” and to recognize this agenda as part of everyone’s job.
Today, many people view corporate profits as dis- tasteful or shameful but, in fact, it is the generation point of jobs. Beyond the pure profit motive, companies have a responsibility for the welfare of their employees, another aspect in the well-designed Six Sigma program. If we inte- grate the philosophies of “profit motive” and “employee relations” into a single point of view, then we have a pow- erful and potentially revolutionary change for the better